by Osteoporosis
Agent Skill and repository template for end-to-end software development entirely inside ordinary Web AI chat.
# Add to your Claude Code skills
git clone https://github.com/Osteoporosis/luna-chat-coderGuides for using ai agents skills like luna-chat-coder.
luna-chat-coder is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by Osteoporosis. Agent Skill and repository template for end-to-end software development entirely inside ordinary Web AI chat. It has 59 GitHub stars.
luna-chat-coder's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/Osteoporosis/luna-chat-coder" and add it to your Claude Code skills directory (see the Installation section above).
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh luna-chat-coder against similar tools.
No comments yet. Be the first to share your thoughts!
Unlocks once the catalog security scan passes (runs nightly).
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
Version 0.1.3
Use ordinary ChatGPT Web conversations for real GitHub repository work—without running a local coding agent, opening a tunnel, or giving the chat access to your computer.
ChatGPT already has a sandbox that can run code. The catch is that network restrictions can stop repository work when the chat needs source, dependencies, or a reliable way to publish a larger change. Luna teaches the model to keep the development loop in that built-in sandbox and use connected GitHub access only for the missing pieces.
The point is simple: give the chat a repository and a development task, not a new piece of infrastructure to operate.
For the ChatGPT Web setup documented here:
That is the normal workflow. A repository created from this template already contains Luna, and you should not need to mention Luna by name or manage its internal recovery steps yourself.
Organization policy may require an administrator to approve the Plugin or GitHub App.
Luna reads the repository's own instructions and requirements, recovers the exact source it should work from, and uses the chat sandbox for the normal edit/test loop.
When direct sandbox access is not enough, Luna can use the connected GitHub path for the missing step. If that path still cannot complete the step reliably, a bounded GitHub Actions run can handle it, then return the work to the sandbox when possible. GitHub Actions is not the default coding environment.
Copy the complete skill directory:
.agents/skills/luna-chat-coder/
Then merge the short Luna entry-point instruction from AGENTS.md into the repository's existing agent instructions. Keep the project's own engineering guidance; Luna works around it rather than replacing it.
For the documented ChatGPT Web path, connect the GitHub Plugin and grant the ChatGPT Codex Connector access to the repository before asking the chat to work on it.
Runtime behavior lives in SKILL.md. Operational details are in actions-missions.md and recovery.md. design-rationale.md is maintainer memory for changing Luna itself; normal skill use does not depend on it.
Luna follows the Agent Skills structure. ChatGPT Web is the path documented and tested here; another host can use the same skill when it provides equivalent sandbox and GitHub capabilities.
MIT. See LICENSE.